Micron Technology (MU.US) closed out its fiscal 2026 with a near-flawless earnings report after the bell on Thursday, Eastern Time. The results showed fourth-quarter revenue of $54.23 billion, up 379% year over year and 31% sequentially, marking a record for the sixth consecutive quarter; adjusted earnings per share of $33.42, more than a tenfold increase year over year; and gross margin climbing to 87%, up 2.1 percentage points from the prior quarter.

For the full fiscal year, Micron's total revenue reached $133.19 billion, 3.5 times the record level of the previous fiscal year, with data center revenue quadrupling year over year and full-year DRAM revenue surpassing $100 billion for the first time. Guidance for the next fiscal quarter also far exceeded expectations: revenue guidance midpoint of $61.5 billion, above Wall Street's $57.57 billion estimate; adjusted EPS guidance midpoint of $38.15, above the $35.81 consensus.

Yet after the report was released, Micron shares initially rose about 2% in after-hours trading before quickly turning negative, seesawing between bulls and bears. As of the latest reading, the stock was trading around $1,060 in pre-market, down roughly 0.5%. Why did a nearly perfect earnings report fail to ignite market enthusiasm?

Strength Beyond the Numbers

The strength of Micron's report is first reflected in the qualitative shift in its revenue mix. In the fourth fiscal quarter, core data center business unit (CDBU) revenue reached $18 billion, up 56% sequentially, with gross margin as high as 90% and a more than tenfold year-over-year increase, driven primarily by AI server demand.

The data center business now accounts for 33% of total revenue, while cloud storage revenue was $16.3 billion, or 30%, bringing the combined total to more than 60% of overall revenue. In DRAM, revenue reached $39.8 billion, up 343% year over year, with bit shipments rising by a mid-single-digit percentage sequentially but prices climbing by a high-single-digit percentage, indicating that growth was driven mainly by pricing rather than pure shipment volume, reflecting extremely tight industry supply and demand.

NAND revenue was $14.1 billion, up 526% year over year and 42% sequentially, with bit shipments up about 10% and prices up about 30%, similarly confirming the severity of supply bottlenecks. Also noteworthy is that Micron disclosed key information about future capacity on its earnings call: more than 75% of fiscal 2027 shipments have already been locked in through long-term agreements, the number of strategic customer agreements (SCAs) rose from 16 in the prior quarter to 26, total customer prepayments increased from $22 billion to $32 billion, and remaining performance obligations (RPO) reached approximately $150 billion, a sharp increase from about $100 billion in the prior quarter.

These agreements are expected to cover more than 35% of Micron's revenue through 2030, with about three-quarters having established pricing frameworks, most with price ranges, and some extending as far as 2031. On capital expenditure, Micron announced about $25 billion for the first half of fiscal 2027, including roughly $11.5 billion in the first quarter, with further increases in the second half, most of the incremental spending earmarked for new wafer fabs rather than simple equipment purchases.

On the capacity rollout timeline, first output from the Idaho fab has been pulled forward to mid-CY27, the Singapore fab is expected to produce in 2HCY28, cleanroom completion at the Hiroshima fab has been moved up to late CY28, and construction has broken ground on the New York fab. CFO Mark Murphy revealed on the call that the fourth fiscal quarter generated $44 billion in operating cash flow and $33.2 billion in free cash flow, and pledged to return 100% of excess cash to shareholders going forward.

Memory Is Defining the Boundaries of AI

Micron CEO Sanjay Mehrotra framed the current industry transformation on the call as: "Superintelligence is creating the most compelling opportunity in Micron's history." That statement is not empty rhetoric. From an industry logic perspective, the scaling of AI models brings not only computing demand but also exponential growth in demand for memory capacity and bandwidth.

Mehrotra noted on the call that running AI applications on platforms with stronger memory capabilities enables more scalable growth and improves end-user experience. In other words, memory is not just an important player in AI infrastructure but also a key variable determining the upper limit of AI system capability. In the HBM space, Micron has partnered with Nvidia (NVDA.US) to develop the industry's first customized HBM solution.

In the first quarter of 2026, Micron began mass production of HBM4 12-high stacked 36GB products for Nvidia's Vera Rubin platform, achieving bandwidth exceeding 2.8TB/s. Mehrotra disclosed that most of 2027 HBM supply has already been contracted, with "prices rising significantly year over year, narrowing the gross margin gap between HBM and conventional DRAM." This means the HBM business is shifting from strategic investment to profit contribution.

Looking longer term, Mehrotra outlined a new growth pole in "physical AI" on the call. He emphasized that L4 and above autonomous vehicles typically have memory capacity exceeding 200GB and storage capacity reaching several terabytes, an order of magnitude higher than current L2+/L3 levels; humanoid robots are expected to have similar requirements. He stressed that by the end of this decade, physical AI will become a significant driver of memory demand.

JPMorgan's supply-demand analysis published before the earnings showed that the HBM market will face supply-demand gaps of 20%, 19%, and 16% in 2026, 2027, and 2028, respectively, with cumulative shortages reaching 23 weeks by 2028. Citi's analysis is more aggressive, projecting that the supply-demand gap will widen from -21% in 2027 to -36% in 2028. Deutsche Bank's estimates suggest the DRAM supply-demand gap will further widen in 2027 and 2028, with the market potentially reaching balance only in 2029. These data point to the same conclusion: AI-driven memory demand has entered a phase of structural shortage rather than cyclical fluctuation.

Why the Stock Is Stuck in a Range

Given such strong fundamentals, why has Micron's stock barely moved? First, gross margin guidance became a short-term sentiment brake. The fourth-quarter gross margin of 87% did beat analyst expectations of 86.9%, but next quarter's gross margin guidance of about 86.25% was below the market's 87.4% estimate.

CFO Murphy explained that the first fiscal quarter will be the low point for gross margin in fiscal 2027, mainly due to higher-cost inventory sales and higher compensation expenses, with margins expected to gradually recover in subsequent quarters, though price increases will moderate. With expectations extremely full, even a 0.45 percentage point shortfall in gross margin guidance was enough to trigger profit-taking by some investors.

Second, the sharp jump in capital expenditure overshadowed the positive earnings beat. The market began to worry that Micron is shifting from a "gross margin expansion model" to a "capacity expansion model." Some analysts noted that as financial results continue to beat expectations, Micron's stock may enter a consolidation phase, no longer rising sharply simply on positive earnings. The significant increase in capital expenditure implies higher future depreciation expenses and potential structural pressure on margins, a key factor weighing on the short-term stock price.

Other analysis suggested that the "capex increase" news overshadowed the earnings beat in after-hours trading. Goldman Sachs analyst James Schneider's team also noted in a latest research report that Micron's latest quarterly results and next-quarter guidance were significantly above Wall Street expectations, while the company disclosed more long-term customer agreements and sharply raised its capital expenditure plan, factors expected to support a modest rise in the stock; however, gross margin was slightly below Goldman's expectations, and investor expectations had already been elevated, limiting room for further valuation expansion.

From an investor perspective, and perhaps most fundamentally, the market's focus is shifting from "how strong is the boom" to "how long can the boom last." Morgan Stanley noted after the earnings that the market's focus is shifting, with investors no longer just caring about how good current results are but beginning to examine whether this supercycle can persist beyond 2028.

Micron's stock has already risen substantially this year, and the market has priced in the earnings beat quite fully. Despite strong fundamentals, Micron still trades at about 7 times forward earnings, well below the roughly 10 times average of the past two years, with expectations of a sharp slowdown in fiscal 2027 and 2028 growth already partly reflected in the valuation.

Moreover, expectation management may have even entered a phase where "beating expectations is not enough." Some analysts noted that analysts have come to treat Micron beating its own guidance as the norm, and the market expects a "beat of a beat." Options market data showed that over the past 10 earnings reports, Micron's stock moved an average of 9.4% the next day, with a median of 9.1%, but this time options priced in only about a 6.3% move, indicating that expectations for sharp post-earnings volatility had clearly diminished.

Goldman Sachs maintained a "neutral" rating on Micron after the earnings but raised its target price from $1,100 to $1,250, based on 18 times earnings and an upwardly revised normalized EPS of $70. Schneider's team concluded that the risk-reward at current levels is roughly balanced, but would consider a more positive view on Micron if the industry shows sustained supply discipline in 2028 and beyond.

In fact, the significance of Micron's earnings report goes beyond a single company's performance. Its core signal is that AI infrastructure buildout's demand for memory is not a short-term pulse but a long-term structural trend that could last until 2028 or even longer. For investors, however, the challenge is that when a company's strong fundamentals have become consensus, the source of excess returns is no longer confirming the boom itself but judging whether the cycle's durability can exceed market expectations. The current valuation of about 7 times forward earnings both reflects market concerns about slowing growth and may contain undervalued long-term value.